Real Time Search: Opportunity or Hype?
Real-time search has been getting so much buzz lately. But is it really everything marketers hoped for? What does it even mean? Let’s find out.
Moderator: Danny Sullivan, Editor-in-Chief, Search Engine Land
Gerry Campbell, CEO, Collecta
Jeremy Hylton, Software Engineer, Google Inc.
Ken Moss, CEO & Co-founder, CrowdEye
Kimbal Musk, CEO, OneRiot
Vipul Ved Prakash, Co-founder, Topsy
Danny’s starting us off by answering the question “What is real-time search, really?”
Web search = search for Web pages
Image search = search for images
News search = search for news
Real time search = search for real times?
Real time content = content published within seconds of being found/created/thought.
Some examples are pics posted within seconds of being taken, or a tweet or status update.
Real time search = finding microblogged content
You can do real-time searches on Twitter. The disadvantage of it is time is the ranking mechanism, and they have a bad habit of losing tweets.
Facebook is another real-time search engine. But it only searches in Facebook and delivers results from people who opted in.
There are also a number of Meta real-time search engines. They tend to let you search for real-time sources among multiple sources. A lot of them are predominately designed to tell you what’s being shared in real-time, not necessarily the current thought process. Two real-time search engines can have radically different results. Until Twitter teams up with a major player, we’re not going to have a great real-time search experience for tweets.
Kimbal Musk steps up. He types “SMX” into OneRiot.com. Results are about the keynote this morning and some of the other session coverage. But in Google, it’s the SMX site and other results that aren’t about what’s happened at the show so far. So if you’re looking for what’s being talked about at SMX right now, you need to use real-time search.
Real-time search is starting to be leveraged at properties across the Web. The real-time search results are starting to appear in many places. It’s a real and growing market.
Jeremy Hylton is next and says that when users have an information need, freshness and real-time is a part of that. Freshness has been an interest for Google for a while. Over time Google has become faster at uncovering new content.
There’s lots of real-time content available, but it’s not necessarily of use to a searcher. If we can take timely content and deliver it to users in context when they need it, that’s going to be very useful for users.
But you still have to worry about spam and duplication. And you need good ranking — we have to be aware of international issues like language, and knowing where the user is if they’re looking for something they just saw. One worry is how open these platforms are and how accessible your content is. We need to be careful about letting the actual authors control how it’s indexed and discovered.
Ken Moss takes the mic and says that real-time search helps consumers and businesses keep their finger on the pulse of the real-time Web. Microblogs are a big deal. Social conversations are a new mode of human communication.
There are some weak aspects of real-time search that are being addressed with new services.
- Perspective: knowing the history of a search term or topic — hours ago, days ago
- Ranking: finding the most relevant results — not just about time
- Topics: a tag cloud can help the user understand what the conversation is about
They’re also experimenting with real-time searches as a dashboard or home page.
Gerry Campbell is up next. Real-time search means so many different things. Is real-time search overhyped? He says it absolutely is. Not because there’s not a ton of opportunity, but because no one’s certain of what the opportunity is yet. It’s here to stay but we’re just at the beginning.
How do we know it’s got value? Don’t measure it by the press; measure it by user intent. People must stay up on their brand and the conversations around them. Microblogging is just the beginning of real-time utterances — it’s a bellwether that people are looking to express themselves. They’ve found about 20 percent of microblogging is commercial. But the query stream is not stable. Though, trying to predict user intent over time is relatively stable.
Vipul Ved Prakash is our next presenter. At Topsy they are indexing all the links that people are posting in real-time formats. The search results page takes the signals shared on Twitter and culls it for the top results. As you go back in time, you start discovering more canonical documents — pages you’d find on a traditional search engine. As the data grows, we’ll have to find ways of condensing it and culling it to find the signal.